Side-by-Side Artificial Intelligence Model in SAP Systems Against Cloud Dependency

To reduce cloud dependency and protect data sovereignty in SAP and artificial intelligence integration, a side-by-side model operating in a separate architecture without altering traditional systems comes to the fore. According to Red Hat data, a significant portion of German companies foresee risks in changing artificial intelligence providers.

While SAP has increased artificial intelligence integration in recent years with S/4 HANA versions and cloud-focused solutions, CEO Christian Klein states that two-thirds of new contracts include artificial intelligence components. However, many existing customers cannot migrate to the cloud due to the complexity, cost, and data sovereignty concerns of on-premises SAP ECC environments. According to Red Hat data, while 57 percent of German companies have an exit strategy, 37 percent of these expect an impact on business continuity if they change providers.

For artificial intelligence use in traditional SAP environments, a side-by-side architecture that does not change the core system is preferred. In this model, artificial intelligence systems run on a separate infrastructure while accessing main SAP data via OData or REST interfaces. Analysis results can be returned in JSON format via APIs and displayed in SAP Fiori applications. Thus, companies can protect data privacy by using open-source or local models such as IBM Granite, LLaMA, and Mistral, and process their data using fine-tuning or RAG (Retrieval-Augmented Generation) methods.